System software concall minutes - 2006-12-19

[Mitch Bradley]

System SW Meeting Minutes
2006-12-19

  Jordan (moderator), Chris Ball, Andres, Jim, Mitch (scribe),  
  Richard, J5, Zephaniah

Status:

Ronak: info about fixes for wireless boot problem supplied,
  forwarded to Marcello, et al

Touchpad: Richard & Andres think that the EC is at fault.
  The EC never pulls down the data line.  Scope traces will be
  sent to Quanta for analysis.

Battery management EC code:
  Getting some bug reports about the A62 EC bits.  Part of the problem
  may be an inadequate battery level measurement chip.  New trac ticket
  entered.

Serial number to vendor source map:
  We now have a spreadsheet correlating serial numbers to which parts
  are installed.  But it is wrong, because it says that all units
  have Suyin camera chips, which we know to be incorrect.

Schedules:
  Current best guess for B2 build start is Jan. 28/29
  Touchpad is the gating item.
  Several people will be there for the bringup (1/5) and
  others for the actual build.

AI: Jim to publish schedule tomorrow.

No new keyboard layouts for B2, at least not yet...

Performance:

Boot time:
  Chris Ball has bootchart running, showing which processes are
  running at which stage of the startup.
  Adam Jackson has been working on X startup.  He has found a
  lot of cruft that can be removed or sped up.
  Some issues with Python that require advice from the experts.
  Cairobench is in the tinderbox now.
  jg thinks that current Cairo and Pango may be an improvement.

We have sysprof now.  jg wants data on X and the browser (rendering);
suspects that rendering slowness may be due to Gecko.

An activity benchmark is available now.

We have some optimized string operations that need to be merged.

QA Processes:
  We need a QA function that is separate from engineering.
  Talk about it later when Blizzard is back.

Mitch:
  Code for device tree via /proc nearly ready
  MSR-setup table for cold boot and resume is complete; speeds up startup
  PCI config spoofing for VSA elimination mostly working

Jordan:
  USB collector bug can be closed
  Formal QA would be nice to give us confidence that it is really fixed